Enhance gene leads tooltip: plot reference expression by tissue

[eweitz]

This shows a bar chart to assess typical tissue-specific gene expression more richly than before (#363).

With this data newly available in Gene Leads, researchers can easily answer questions about genes of interest, e.g.:

• In which tissues is my gene typically most expressed?
• Is expression usually concentrated in one tissue, or across many?
• How does the expression of this gene in this sample compare to that in a large, representative cohort?

User interface

Here's how it looks!

Tissue_detail__Ideogram_2023-12-23.mov

Data architecture

This UI enhancement leverages a new-to-Ideogram technique for data retrieval. Like other gene leads data, this new data on median expression for up to 10 tissues per gene is stored in the service worker cache. However, newly, essentially all of the data is kept on disk and not read into memory. Instead, the Ideogram cache only reads a small "line-byte index" into memory.

The index maps each gene to the byte-offset of the line in the larger tissue-detail cache file on disk. Then, using workbox-range-requests, Ideogram reads only the needed byte range into memory. This is effectively the same as an HTTP range request.

So speed is the same as before, but memory footprint is drastically smaller.